arXiv:2606.27629v1 Announce Type: cross Abstract: Cross-platform deployment of offensive comment detection for Chinese social media suffers performance degradation. The paper proposes a dual-threshold hard mining method to address this. First, the clean-Chinese-base RoBERTa is finetuned on COLD to establish a binary baseline for fair comparison. Second, a three-class fine-labeled test set covering Weibo, Xiaohongshu, Tieba, and Zhihu is constructed, domain distances from the source are quantified using Jaccard and Proxy-A Distance, as well as the degradation bottleneck of the baseline under do
The proliferation of Chinese social media platforms and the increasing need for effective content moderation across them drives the development of advanced detection methods.
Improving cross-platform offensive content detection is crucial for maintaining platform integrity, managing online discourse, and potentially influencing regulatory frameworks for AI-powered moderation.
The ability to more effectively detect and mitigate offensive content across diverse Chinese social media platforms improves the robustness and reliability of AI moderation systems in complex linguistic and cultural contexts.
- · Chinese social media platforms
- · AI content moderation companies
- · NLP researchers
- · Online users
- · Malicious online actors
- · Unmoderated content creators
Improved content moderation leads to a cleaner and potentially safer online environment on Chinese social media.
The techniques developed could be adapted for cross-platform content moderation in other languages or for different types of undesirable content.
Enhanced AI moderation capabilities might fuel further discussions and regulations around censorship, freedom of speech, and algorithmic bias in content control.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI